Abstract In research on referring expressions (REs), multiple explanations have been proposed for second language (L2) overexplicitness—the phenomenon of intermediate‐to‐advanced learners producing fuller REs than pragmatically required. However, these explanations have rarely been empirically tested. This study examined the cognitive load hypothesis and its interaction with the error avoidance and clarity hypotheses using a mixed‐model quasi‐experimental design. Sixty‐nine Chinese learners were assigned to either a language‐focused or content‐focused group and completed three retellings of one narrative. Cognitive load was manipulated through task repetition and content familiarity, whereas the avoidance and clarity hypotheses were operationalized via task instructions targeting accuracy and communicative clarity, respectively. Results show that task repetition significantly reduced overexplicitness, largely supporting the cognitive load hypothesis. In contrast, no significant effects were observed for the other hypotheses, suggesting overexplicitness was not driven by deliberate strategies of avoidance or hyperclarity. These findings indicate effective means of fostering pragmatically appropriate referential choices.
Ryan et al. (Fri,) studied this question.